首页 | 本学科首页   官方微博 | 高级检索  
     

基于TLD框架的上下文目标跟踪算法
引用本文:吕枘蓬,蔡肖芋,董亮,涂继辉.基于TLD框架的上下文目标跟踪算法[J].电视技术,2015,39(9):6-9.
作者姓名:吕枘蓬  蔡肖芋  董亮  涂继辉
作者单位:1. 武汉大学测绘遥感信息工程国家重点实验室,湖北武汉,430079
2. 武汉大学测绘遥感信息工程国家重点实验室,湖北武汉430079;长江大学电信学院,湖北荆州434023
基金项目:高分辨率对地观测系统重大专项;国家高技术研究发展计划(863计划)
摘    要:提出了一种基于TLD (Tracking-Learning-Detection)框架的上下文目标跟踪算法.在TLD框架中,融入时空上下文跟踪算法,提高跟踪器的鲁棒性和稳定性.引入Kalman滤波来处理目标被严重遮挡时跟踪失效的问题.此外,采用由粗到精的搜索策略进行目标检测,利用帧差法确定运动目标疑似区域,提高检测效率.实验结果表明所提出的算法具有较好的鲁棒性和实时性.

关 键 词:目标跟踪  TLD  时空上下文
收稿时间:2014/11/7 0:00:00
修稿时间:2014/11/23 0:00:00

Context object tracking algorithm based on TLD framework
Lv Ruipeng,Cai Xiaoyu,Dong Liang and Tu Jihui.Context object tracking algorithm based on TLD framework[J].Tv Engineering,2015,39(9):6-9.
Authors:Lv Ruipeng  Cai Xiaoyu  Dong Liang and Tu Jihui
Affiliation:State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University,State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University,State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University,State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University
Abstract:A spatio-temporal context object tracking algorithm based on the framework of TLD (Tracking-Learning-Detection) is proposed. The tracking algorithm via spatio-temporal context learning is worked as the tracker to improve the stability and the robustness in object tracking. Kalman filter is introduced to deal with heavy occlusions. Additionally, frame difference method is employed to detect the candidate region of moving object in the detection module. This coarse-to-fine search strategy reduces the computational complexity in the detection procedure. According to the experimental results, the proposed method has good robustness and efficiency.
Keywords:Object tracking  TLD  Spatio-temporal context
本文献已被 万方数据 等数据库收录!
点击此处可从《电视技术》浏览原始摘要信息
点击此处可从《电视技术》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号